An Improved Adaptive Centroid Estimation Algorithm
Chun‐Yu Chen, Maoliu Lin · 2006
In this paper, we study the statistic characteristics of centroid of noisy signal firstly. Because of additive noise and random relative shifts among signals, the centroid of noisy signal is a random variable. We obtain analytical form of the variable, which is composed of power ratio of signal to noise, centroid of noise and centroid of signal. So from the statistic characteristics of centroids of noise and signals, we get the one of noisy signal. We analyze the principle of centroid estimation algorithm and conclude that the effective way to decrease the distortion of signal estimated is to increase the signal-to-noise ratio (SNR) before alignment. Secondly, we study the adaptive centroid estimation (ACE) algorithm and present an improved version. The simulations indicate that it has better performance under the condition with lower SNR. Finally, we apply these two algorithms on "nose-to-nose" calibration technique, and the signals to be aligned are collected from experiments. The result shows that the uncertainties of phase of "kick-out" pulse estimated by improved ACE algorithm is less than the one estimated by ACE algorithm